in vivo drug discovery in the zebrafish

R E V I E W S

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | JANUARY 2005 | 35

Drug discovery involves a complex iterative process ofbiochemical and cellular assays, with final validation inanimal models, and ultimately in humans. Mammalianmodels of absorption, distribution, metabolism andexcretion (ADME)/pharmacokinetics and efficacy areexpensive, laborious and consume large quantities ofprecious compounds. There is also increasing pressureto limit animal use to situations in which they areabsolutely necessary, such as in preclinical toxicity andsafety assessment. Zebrafish are beginning to be used atvarious stages of the drug discovery process and can bea useful and cost-effective alternative to some mam-malian models (such as rodents, dogs and pigs). In thisarticle, we will review the use of zebrafish in target vali-dation, disease modelling, target and lead compounddiscovery, and toxicology.

Zebrafish backgroundThe zebrafish has been a popular pet for decades. Its usefor research increased substantially approximately 8years ago, following the demonstration that it isamenable to large-scale forward genetic screens1. Suchlarge-scale genetic screens had previously been limitedto invertebrates such as flies, worms and yeast. Zebrafishmade it possible to take a forward genetic approach tounderstanding vertebrate-specific processes that affectdevelopment and disease. In the past decade, thou-sands of mutations have been generated that affect

ORGANOGENESIS, physiology and behaviour2,3. These muta-tions have proved to be a rich source of informationabout the relationships between genes and function.

In the past few years, several additional tools havebeen developed that have greatly increased the utility ofthe zebrafish as an experimental organism (TABLE 1). Thezebrafish genome has now been sequenced by theSanger Center, and there has been substantial anno-tation of the genome through the trans-NationalInstitutes of Health Zebrafish Genome Initiative.Collections of full-length zebrafish cDNAs are available,as are multiple DNA microarrays for expression-profilingexperiments4,5. Gene expression can also be rapidlyanalysed using whole-embryo in situ hybridization.Gene function can be rapidly and robustly studied inzebrafish using antisense MORPHOLINO OLIGONUCLEOTIDES6,7.Furthermore, techniques for generating transgeniclines8–10, targeted mutations (reverse genetics)11 andcloning by nuclear transfer12 have been developed.Together, these tools have made the zebrafish the organ-ism of choice for many researchers, as demonstrated bythe marked increase in zebrafish publications in recentyears. The yearly number of PubMed references onzebrafish has almost tripled in the past 5 years andincreased more than tenfold in the past decade (FIG. 1a).

Beyond genetics and experimental tools, the strengthof the zebrafish resides in the analysis of phenotype.Perhaps no other organism (and certainly no vertebrate)

IN VIVO DRUG DISCOVERY IN THEZEBRAFISHLeonard I. Zon* and Randall T. Peterson†

Abstract | The zebrafish has become a widely used model organism because of its fecundity,its morphological and physiological similarity to mammals, the existence of many genomictools and the ease with which large, phenotype-based screens can be performed. Because of these attributes, the zebrafish might also provide opportunities to accelerate the process ofdrug discovery. By combining the scale and throughput of in vitro screens with thephysiological complexity of animal studies, the zebrafish promises to contribute to severalaspects of the drug development process, including target identification, disease modelling,lead discovery and toxicology.

*Howard Hughes MedicalInstitute, Division ofHematology/Oncology,Children’s Hospital, HarvardMedical School, Boston,Massachusetts 02115, USA.†Developmental BiologyLaboratory, CardiovascularResearch Center,Massachusetts GeneralHospital, Harvard MedicalSchool, Charlestown,Massachusetts 02129, USA.Correspondence to R.T.P.e-mail: [email protected]:10.1038/nrd1606

ORGANOGENESIS

The formation of organsduring ontogeny.

MORPHOLINO

OLIGONUCLEOTIDE

A synthetic oligonucleotidewith a backbone that has beenchemically modified to resistnucleotide degradation.

MAUTHNER CELLS

A bilateral pair of brain stemneurons that receive sensoryinputs and trigger thecharacteristic escape responseof fish to aversive stimuli.

36 | JANUARY 2005 | VOLUME 4 www.nature.com/reviews/drugdisc

R E V I E W S

efforts directed at poorly validated targets frequently fallshort of expectations16,17.

Genetic and morpholino oligonucleotide screensin the zebrafish are an efficient means of systemati-cally assessing the roles of individual genes in diseaseprocesses. As such, they represent a promising routeto the identification and validation of novel drug tar-gets. In traditional forward genetic zebrafish screens,phenotypes have been identified that resemblehuman diseases18–20. The novel genes that underlie thezebrafish disease phenotypes might lead directly tothe identification of novel drug targets. Alternatively,the disease phenotypes might form the basis of fur-ther screens to find genes or chemicals that can correctthe phenotype.

Genetic screensDuring the past decade, numerous forward geneticscreens have been carried out in zebrafish. Thousandsof distinct mutations have been identified, and morethan 400 of them have been cloned2,3,21. So far, theemphasis of most of these screens has been on devel-opmental biology. Therefore, although these screenshave provided a tremendous amount of informationabout the roles of genes during embryogenesis, therelevance of most of the identified genes to diseasepathology remains poorly established. However, severaldisease-related screens have been performed, includingthe following examples.

Polycystic kidney disease. Forward genetic screens havebeen used to identify several mutations that cause kidneycysts in zebrafish analogous to those found in humanswith polycystic kidney disease (PKD)22,23. The mutatedgenes include pkd2 and vhnf1 (also known as transcriptionfactor 2; tcf2), genes that are implicated in human PKD,

is better suited to high-throughput phenotyping. Thezebrafish embryo is optically transparent, making itpossible to detect functional and morphologicalchanges in internal organs without having to kill ordissect the organism (FIG. 1b–e). These functional andmorphological changes can be further highlighted bythe use of transgenic lines and other reporter molecules(FIG. 1f). For example, fluorescent labelling of MAUTHNER

CELLS was recently used to identify conditions that pro-mote post-injury regeneration of damaged zebrafishneurons13. In another example, fluorescently quenchedphospholipids were used to detect lipid processing andtransport in live, intact zebrafish14. The transparent fishand fluorescent marker were combined to facilitate therapid phenotyping of thousands of individuals in alarge-scale screen for genes that affect lipid trafficking.

The scale that can be achieved in zebrafish experi-ments is impressive by vertebrate standards. Earlyzebrafish embryos are less than 1 mm in diameter,allowing several embryos to fit easily in a single well of a384-well plate (FIG. 1g). What they lack in size they makeup for in numbers — each female can lay up to 300 eggsat a time. Even a small zebrafish facility can generatemany thousands of embryos at a time. Miniature size,large numbers and easy detection of a phenotype ofinterest are the sine qua non of high-throughput screen-ing (HTS) and, indeed, zebrafish have proved to be wellsuited to HTS, as described in FIG. 2.

Identifying novel drug targetsIdentifying and validating novel drug targets remains abottleneck in drug discovery15. Despite a wealth ofinformation about normal physiology and diseasepathology, it is still difficult to predict which targets willeffectively reverse a disease phenotype without producingunwanted side effects. Consequently, drug discovery

Table 1 | The zebrafish toolbox

Technology Description References

Forward genetics

Chemical mutagenesis High mutation rates, large-scale screens 72

Insertional mutagenesis Efficient cloning of mutations 73

Reverse genetics

Morpholinos Rapid, inexpensive gene knock-downs 6

TILLING Directed identification of permanent mutations 11

Expression profiling

Gene chip Zebrafish Affymetrix chip 77

Spotted microarrays cDNA and oligonucleotide microarrays 4,5

Other tools

Transgenesis Rapid production of stable transgenic lines 8–10

cDNA collections Full-length cDNA collections 78

Mutant collections Thousands of catalogued mutant linesHundreds of lines available through public stock centres 79,80

Physical and genetic maps Radiation hybrid and microsatellite genetic linkage maps 74–76

Genomic sequence 5.7-fold coverage of the zebrafish genome 81,82Substantial genome annotation

TILLING, targeting-induced local lesions in genomes.


R E V I E W S

as well as several novel genes. Many of the identifiedgenes are involved in cilia function, lending strong sup-port to the emerging hypothesis the PKD is caused byciliary defects.

Cholesterol processing. A genetic screen has been usedto identify zebrafish mutants with normal digestiveorgan morphology that have altered phospholipidand cholesterol processing14.

Tissue regeneration. Mutants have been identified thataffect the ability to regenerate damaged tissues, includingfins and hearts24.

Heart disease. Mutants identified through the use of zebrafish screens recapitulate several aspects ofhuman heart disease. In several cases, the zebrafishand human disease states are caused by orthologousmutations. In both zebrafish and humans, TTN (titin)mutations cause cardiomyopathy25,26; TBX5 (T-box 5)mutations cause congenital heart defects27,28; andHERG (also known as potassium voltage-gated chan-nel, subfamily H (ether a go go-related), member 2;KCNH2) mutations cause arrhythmias29,30.

Anaemias. The iron transporter ferroportin was firstdiscovered in zebrafish as a mutant with embryonicanaemia31. Subsequently, it was found that ferroportin isdefective in patients with HAEMOCHROMATOSIS TYPE 4 (REF. 32).This is the first time that a zebrafish gene has led to thediscovery of a new human disease gene. Also, the dis-covery of the kgg mutant, which lacks haematopoieticstem cells and is defective in the cdx–hox pathway, led toa new method for amplifying blood stem cells frommammalian embryonic stem cells33.

Cancer. It is possible to model cancer in the zebrafishusing various transgenic fish. So far, there are excel-lent models of leukaemia and melanoma in thezebrafish20,34,35.

Nervous system. Zebrafish models of numerous ner-vous system disorders have been described, includingretinal degeneration and anxiety36,37. In addition,genetic screens to identify mutations that cause deaf-ness, blindness, mechanosensation defects, loss ofCONDITIONED PLACE PREFERENCE and other behaviouraldefects have been reported38–41.

The forward genetic screens and models describedabove provide important entrance points into path-ways that might contribute to numerous diseaseprocesses. Ultimately, genetic suppressor screens arelikely to add great value. For example, starting with adisease model, it should be possible to screen formutations that suppress or delay the development ofthe disease phenotype, with the mutated genes becom-ing logical drug target candidates. A similar approachin mice was recently reported, in which genetic suppres-sors of genetic THROMBOCYTOPAENIA were identified. Twomutant alleles of c-Myb were identified that suppressthe disease phenotype, indicating that c-Myb might be

Ann

ual z

ebra

fish

Pub

Med

list

ings

0

100

200

300

400

500

600

700

800

900

1980 1985 1990 1995 2000 2003

a

b c

d e

f g

Figure 1 | The zebrafish as a system for biomedical research. a | The increase in the use ofzebrafish for research is depicted as the number of zebrafish-related Pubmed references foreach year from 1980 to 2003. The term ‘zebrafish’ was used for searching Pubmed. b–e | Thetransparent body of the zebrafish larva is shown 5 days post-fertilization. Images of increasingmagnification show the ability to discern structures on the systemic (c), organ (d, heart shown)and cellular (e, tail shown) scales in the living organism. f | Specific structures can behighlighted in the living zebrafish by transgenic expression of fluorescent proteins from tissue-specific promoters. Here, endothelial cells are marked by expression of enhanced greenfluorescent protein from the fli1 (friend leukemia integration 1) promoter. g | Two zebrafishembryos developing in a single well of a standard 384-well assay plate.


R E V I E W S

specific mRNAs6. When injected into zebrafish embryos,they cause a robust knockdown of gene function. Thedevelopmental roles of many individual genes have beendetermined by injection of morpholinos in zebrafish.

Could morpholino oligonucleotide-mediated geneknockdown be used to discover novel therapeutic targets?One promising approach involves inducing a diseasestate in zebrafish (through the use of infectious agents,genetic mutations or pharmacological agents) com-bined with large-scale morpholino screening. By system-atically knocking down many genes, it should be possibleto identify gene knockdowns that prevent or slow thedevelopment of the disease phenotype. Small-scalemorpholino screens have been reported in Xenopustropicalis43 and Ciona intestinalis44, and zebrafishlibraries containing morpholinos that target severalhundred genes have been assembled. As the size andavailability of these libraries increase, they are likely to bepowerful tools for the systematic, unbiased identificationof therapeutic targets.

Therapeutic targets identified by forward genetic ormorpholino oligonucleotide screens could become thefocus of conventional, target-based drug discoveryefforts, including in vitro HTS. Alternatively, small-molecule screens to identify suppressors of zebrafish dis-ease phenotypes can be initiated before the identificationof a validated target, as described below.

Phenotype-based target and lead discoveryWhen a therapeutic target has been identified and vali-dated, in vitro HTS based on target binding or functioncan often be used to identify novel structures thatmodify the activity of the target protein. Alternatively,medicinal chemistry can frequently be used to generatenovel derivatives of compounds that are known toaffect the target. For example, since the discovery of the3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA)reductase inhibitor mevastatin, dozens of structurallyrelated statins have been developed45.

What about diseases for which the optimal thera-peutic targets have not yet been identified? At present,validated targets represent only a small fraction ofpotential targets, and developing therapies for many ofthe most significant diseases (including atherosclerosis,chronic myelogenous leukaemia and bronchoalveolarlung cancer) is limited by the fact that effective targetshave not yet been identified for these diseases. Predictingwhich proteins must be targeted to improve a diseasestate remains surprisingly difficult and, consequently,selecting novel targets for drug development on thebasis of in vitro data entails significant risk16,17. In manycases, drug development directed at novel targets hasproceeded smoothly at early in vitro stages but has failedat late stages, either because disruption of the target didnot produce the expected effect in whole animals orbecause it also caused unexpected side effects. Clearly,many in vitro enzymatic assays are poor surrogates forcomplex physiological diseases. It is perhaps not surpris-ing, therefore, that developing novel therapies guidedsolely by in vitro assays often leads to unexpected resultsin whole organisms.

a therapeutic target for thrombocytopaenia42. Geneticsuppressor screens in zebrafish have not been reported,but all of the required elements for such screens arenow available.

Morpholino oligonucleotide screensMorpholinos are chemically modified antisense oligo-nucleotides that can be designed to hybridize to thetranslation-initiation or splicing acceptor/donor sites of

HAEMOCHROMATOSIS TYPE 4

A hereditary condition that ischaracterized by the abnormalaccumulation of iron in tissuesof the body.

Malezebrafish

Femalezebrafish

a

b c

d

e

f

N

S

SHO

O

O

S S

NO COOH

NH

HN

O

O

CH3

N

O

O

O

HN

HN

HN

SClCl

Cl

ON

SHN

N

Br

O

O

Cl

HN

NH

O

O

O

HO

N

O

NH2

H3C

O2N

CH3H3C

O O CH3

CH3ONO2

NN

O

OOH

O2N

NO2

OOH

OHN

NHO O

HN

O

HN

Cl ClCl

HN

SNH

O

CH3NS S

NCOOHO

S

SOH O

O

OH

O

HO

O

Figure 2 | Zebrafish small-molecule screens. a | Adult zebrafish are mated, producing 200–300embryos per female. b | Embryos are distributed to 96- or 384-well assay plates. Typically, threeembryos are placed in each well. c | A small-molecule library that contains potentially biologicallyactive compounds is synthesized or acquired. d | Small molecules are added to the watersurrounding the zebrafish. A single small molecule or combinations can be added to each well.e | After a period of incubation, the phenotypic effects of the small molecules on the zebrafishare determined visually or through an automated read-out. f | Referral to the library databasereveals the identities of the biologically active small molecules.


R E V I E W S

compounds, take up less space and use simpler pheno-typing than the mammalian assays described above.All of these requirements can be met by screens inwhole zebrafish. The feasibility of performing pheno-type-based small-molecule screens in zebrafish isillustrated by two recent examples in which chemicalsuppressors of specific genetic mutations were identified.Zebrafish that have the gridlock mutation, whichcauses a vascular defect, and crb (crash and burn), acell-cycle mutation, were used to screen throughdiverse small-molecule libraries. In both screens, smallmolecules that can reverse the phenotypic effects ofthe mutations were identified49,50.

A screen for suppressors of the gridlock mutation.Gridlock mutants have a hypomorphic mutation in thehey2 gene, a hairy/enhancer of split-related transcrip-tional repressor believed to function downstream ofNotch in regulating vascular cell-fate decisions51,52.Mutant embryos develop a dysmorphogenesis of thedorsal aorta that prevents circulation to the trunk andtail, although perfusion of the head is normal. Mutantembryos were distributed into 96-well plates, with threeembryos per well, and compounds from a structurallydiverse small-molecule library were pin-transferred intothe water surrounding the zebrafish embryos, with onecompound per well. The embryos were allowed todevelop and visually scored for the gridlock phenotypeunder a dissecting microscope. After screening 5,000small molecules, two structurally related compoundswere identified that completely suppress the gridlockmutant phenotype, causing mutant embryos that areexposed to the compounds to develop a normal vascu-lature (FIG. 3)49. The gridlock-suppressing compounds,called GS4012 and GS3999, represent a novel class ofcompounds that were not previously known to influencevasculogenesis or angiogenesis. The compounds seem torescue the aortic defect directly rather than promotingthe formation of collateral vessels. No ectopic vesselshave been observed in treated embryos (R.T.P., unpub-lished observations), which go on to survive throughadulthood. Their mechanism of action is not yet known,although they seem to activate the vascular endothelialgrowth factor (VEGF) pathway, and overexpression ofVEGF is sufficient to rescue the gridlock mutation49.

The perfusion of tissues with blood has a significantrole in many diseases, including myocardial infarction,stroke, diabetes and cancer. Identifying novel classes ofcompounds and novel mechanisms for regulating theformation of blood vessels might have therapeutic appli-cations. As initial evidence that the gridlock suppressorsmight have vasculogenic activity in a mammalian system,GS4012 was found to promote the formation ofendothelial tubules from human cells in an in vitromatrigel assay49.

A screen for a suppressor of the crb mutant phenotype. Anembryo-based small-molecule screen was performed toidentify suppressors of the crb mutant phenotype50. crbmutants have an increased number of mitotic cells, andthe heterozygotes have an increased rate of cancer.A total

One alternative to in vitro target-based drug discoveryis discovery that is guided by phenotype in the contextof a whole organism. Whereas target-based approachescan lead to the discovery of compounds that modify atarget but that might not modify the disease, phenotype-based approaches can be used to discover compoundsthat modify the disease phenotype, regardless of thespecific molecular target. The development of manydrugs in use today was guided by phenotype in wholeorganisms (TABLE 2). For example, to discover the novelcholesterol-lowering drug ezetimibe (Zetia), scientists atSchering-Plough initially used a target-based approach46.They predicted that inhibitors of acyl-coenzyme Acholesterol acyltransferase (ACAT) would lead to areduction in serum cholesterol, and so they developedin vitro assays for screening for ACAT inhibitors.Although potent, specific inhibitors of ACAT wereidentified, none of the compounds was effective atreducing serum cholesterol in the cholesterol-fed ham-ster. Remarkably, in the process of screening throughthe ineffective ACAT inhibitors, a synthetic byproductwas identified that reduced serum cholesterol in thehamsters. Guided solely by phenotype — namelyserum cholesterol levels in the hamster — the byproductwas identified46. Derivatives of the byproduct weresynthesized and tested for efficacy — again guided byphenotype alone — resulting finally in the develop-ment and Food and Drug Administration (FDA)approval of ezetimibe in 2002. The precise target of eze-timibe remains unknown, although it seems to blockcholesterol absorption in the intestine47.

Because the mechanism of action of ezetimibe isdistinct from other classes of cholesterol-lowering drug,such as the statins, it is a powerful additional tool fortreating hypercholesterolaemia. In this case, the target-based approach to the development of ACAT inhibitorswas unsuccessful, whereas the phenotype-based app-roach enabled the discovery of a novel cholesterol-lower-ing drug that functions through a novel and unexpectedmechanism of action.

The examples described above illustrate the utility ofbeing able to follow the phenotypic effects of com-pounds in a physiologically relevant, organismal context.However, systematizing this approach can be difficult.In the case of ezetimibe, a serendipitous observationsparked further experimentation, which necessitated theuse of many animals and painstaking phenotyping46.Without the benefits of serendipity, a large-scale pheno-typic screen could theoretically have been performedbut would have required even more animals and morephenotyping. In one such Herculean screen, Squibbscientists performed a whole-animal, phenotype-basedscreen for antituberculosis agents. They screened 5,000compounds in mice, en route to the successful pheno-type-based discovery of isoniazid48.

Enter the zebrafish. Despite the demonstrated poten-tial of this approach to discover novel, effective therapies,the resources, space and effort required to carry out thistype of screen have prohibited their widespread use. To bepractical, phenotype-based whole-organism screens needto be much less expensive, consume smaller quantities of

CONDITIONED PLACE

PREFERENCE

An experimental procedure to test an animal’s learnedassociation betweenenvironmental cues and thesubjective state produced by adrug. After repeated exposureto a drug in the presence ofdistinctive environmental cues,the animal is tested to see if itwill preferentially movetowards a compartment thatcontains drug-associatedenvironmental cues, even in theabsence of the drug.

THROMBOCYTOPAENIA

A condition characterized by anabnormal reduction in thenumber of circulating platelets.


R E V I E W S

on absorption, toxicity and so on. By performing SARstudies in whole zebrafish, structures can be identifiedthat improve potency without increased toxicity or lossof in vivo efficacy.

Screens in adult zebrafishSo far, the vast majority of zebrafish experiments havebeen performed with embryonic and juvenilezebrafish, because the transparency of the fish is great-est at these stages. Furthermore, the embryos andjuveniles require much less space than adults and fiteasily into multi-well assay plates. A remarkable rangeof biological and disease process can be studied inthese early developmental stages, including diseasestates such as heart failure and arrhythmias that arefrequently associated with adults. However, there are bio-logical processes that are restricted to the adult, and itis becoming clear that the intact adult zebrafish iswell-suited to many of these studies. For instance, wehave been able to study transplant biology in thezebrafish, having accomplished marrow transplantationin irradiated adults53,54. Studies of regeneration andbehaviour have also been carried out in the adult24,55,56.Although the logistics for screening will necessarily bedifferent, it should be possible to carry out chemicalscreens in whole adult zebrafish.

ToxicologyThe rate of innovative pharmaceutical therapies thatreach patients seems to be slowing: the number of newmolecular entities submitted to the FDA has declinedby about half since 1997. In a recent report, the FDApoints to technological deficits in toxicology as one ofthe primary causes of this ‘pipeline problem’, notingthat in many cases, the approaches of the last centuryare still being used to assess this century’s candidates(see online links box). New animal models are neededto test the safety of novel drug candidates, and the FDAreport that an estimated 10% improvement in predict-ing failures before clinical trials would save US $100million per drug in development costs. In addition tooutdated technologies, toxicology frequently suffers bybeing divorced from the drug discovery process —efforts to discover leads and improve their potencyoften occur independently from the assessment oftoxicity57. Some efforts are being made to involve toxi-cology earlier in the drug discovery process, such aseliminating compounds with problematic chemicalmoieties from screening libraries58 or prioritizing leadson the basis of performance in in vitro toxicity assays59.However, much more progress is needed to developbetter animal models for toxicological assessment and toinvolve toxicology earlier in the drug discovery process.

The zebrafish is rapidly gaining acceptance as apromising animal model for toxicology60. The ability toefficiently assess the toxicity of a large number of com-pounds enables whole libraries to be prescreened forpotential toxicity. In this way, compounds with obvioustoxicity can be eliminated from libraries before HTS.Alternatively, preliminary hits from HTS can all betested for toxicity as a means of prioritizing compounds

of 100 pairs of crb heterozygotes generated enoughembryos to screen 1,000 compounds per week at 20 µMeach. The entire DIVERSet E library (Chembridge, SanDiego), which contains 16,320 compounds, was screenedin 16 weeks. Because embryos were generated by matingheterozygous crb mutants, the embryos used for thescreen had a phenotypic ratio of three wild-type embryosto one mutant embryo. Three classes of compound thathad different effects on wild-type and mutant embryoswere identified. Class I included 21 small molecules thatincreased phosphorylated histone H3 (pH3) staining inboth wild-type and mutant embryos, and class II con-sisted of 11 chemicals that decreased pH3 staining in allembryos (R. Murphey, H. M. Stern and L.I.Z., manuscriptin preparation).A third class consisted of one compoundthat specifically suppressed the elevated pH3 staining ofcrb embryos and had little effect on wild-type embryos.Because of its ability to suppress a specific cell-cycledefect in mutant cells without affecting wild-type cells,this compound might be useful as an anticancer agent.

Structure–activity relationshipsThe zebrafish as a system is highly amenable to the studyof structure–activity relationships (SARs). During theabove screen for compounds that alter pH3, severalcompounds were identified that perturb the cell cycle,but do so in both wild-type and mutant embryos. SARshave been established for several of these compounds.As shown in FIG. 4, several derivatives of parent com-pound L were generated and tested for their ability toalter pH3 levels in intact zebrafish. Three compoundsinduce the zebrafish phenotype at concentrations similarto those of the parent compound, four compoundsinduce the phenotype only at a fivefold higher concentra-tion, and three compounds have no activity (R. Murpheyand L.I.Z., unpublished observations).

One advantage of performing SAR studies inzebrafish is that they couple the analysis of bindingaffinity and ADME/toxicity. In a traditional in vitro SARstudy, structural changes that improve potency are pur-sued, but these changes might have detrimental effects

Table 2 | Frequently prescribed drugs discovered by phenotypic effect

Drug Use

Sildenafil (Viagra; Pfizer) Erectile dysfunction treatment

Warfarin Antithrombotic

Digoxin Heart failure treatment

Benzodiazepines Sedative

Glitazones Antihyperglycaemic

Ezetimibe (Zetia; Merck/Schering) Cholesterol reduction

Gabapentin Neuralgia and epilepsy treatment

Opiates Analgaesic

Hydrochlorothiazide Diuretic and antihypertensive

Buproprion (Wellbutrin; GlaxoSmithKline) Antidepressant

Amitriptyline Antidepressant

Isoniazid Tuberculosis therapy

Metformin Antihyperglycaemic


R E V I E W S

metabolic pathways, DNA repair and other relevantphysiological processes are active.

Cardiotoxicity is another area in which a zebrafishassay has the potential to contribute to drug develop-ment. Drug-induced prolongation of the cardiac QTinterval can lead to the fatal arrhythmia TORSADE DE

POINTES, and QT prolongation has become a leadingcause of failure during drug development. Milan et al.have developed an automated, high-throughput assayfor bradycardia in zebrafish embryos, which they haveshown to correlate with QT prolongation in humans68.They tested 100 compounds in the assay and showedthat 22 of 23 drugs known to cause QT prolongation inhumans cause bradycardia in zebrafish. In addition, theassay was able to detect drug–drug interactions that leadto QT prolongation, such as the well-known synergisticinteractions between erythromycin and cisapride andbetween cimetidine and terfenadine. These interactionsresult from the physiological effects of one compoundinfluencing the metabolism of the second compound andcan only be detected in a whole organism. These resultshighlight the value of performing toxicity studies inzebrafish — zebrafish assays can achieve the scale andthroughput of in vitro assays, but they occur in a rele-vant physiological setting in which complex pharmaco-kinetic and pharmacodynamic processes remain intact.

ConclusionsThe zebrafish has been pioneered as a developmentaland genetic tool for the study of organogenesis anddisease. Zebrafish researchers in the basic sciences havetaken advantage of the high degree of genetic and physio-logical similarity between zebrafish and mammals, theirsmall size, large numbers and the fact that phenotypescan be so rapidly assessed in high throughput. These

for further development. Significantly, toxicologicalstudies in zebrafish also require compound quantitiesonly in the microgram or milligram ranges, whereasmammalian assays frequently require a few grams toseveral hundred grams of a given compound.

Most zebrafish toxicity studies so far have focused onenvironmental contaminants, including pesticides. Thezebrafish is not as well established as a model for drugtoxicology, and questions remain about how relevantfish toxicity is to humans. Nevertheless, studies havebegun to show that many toxic responses are well con-served between fish and mammals. Toxic-responsesimilarities between zebrafish and mammals have beennoted for small molecules that cause endocrine dis-ruption, reproductive toxicity, behavioural defects,teratogenesis, carcinogenesis, cardiotoxicity, ototoxicity,liver toxicity and so on61–66.

Several zebrafish assays have been developed specifi-cally to monitor toxicities of significance to drug develop-ment. For example,Amanuma et al. developed a sensitivezebrafish assay for detecting small-molecule-inducedmutagenesis67. Embryos from a zebrafish line that har-bour a plasmid containing the 30S ribosomal subunitprotein S12 (rpsL) gene can be exposed to a test com-pound, after which the plasmid is excised from thezebrafish genome and transformed into bacteria. Wild-type copies of the plasmid produce colonies that areresistant to kanamycin but sensitive to streptomycin,whereas mutant copies of the plasmid produce coloniesthat are resistant to both kanamycin and streptomycin.Therefore, mutation rates can readily be calculated withgreat sensitivity. Although there are other assays formutagenesis, including the Ames test and micronucleustest, this assay offers the advantage of testing com-pounds in a complete organismal context, in which

TORSADE DE POINTES

A form of polymorphousventricular tachycardia that isassociated with prolongation ofthe cardiac QT interval that canlead to sudden cardiac death.

0 2.5 5 7.5 10

Tail

circ

ulat

ion

(%)

0

20

40

60

80

100d

ca

b

Concentration (µg ml–1)

O

S

CH3

N

S

O2NN

Figure 3 | Chemical screens in whole, mutant zebrafish can identify small molecules that suppress genetic defects.Fluorescent micro-angiograms show the circulation pattern in gridlock mutant embryos (a) or mutant embryos treated with a gridlocksuppressor compound (b). Mutant embryos treated with the gridlock suppressors (c) are rescued in a dose-dependent manner (d).


R E V I E W S

could be several reasons for this, including the factthat cancer cell lines with aberrant cell-cycle regula-tion were used and potential differences between thezebrafish and human cell cycles. However, it is alsopossible that a physiologically relevant organismalcontext is required for the mechanisms of action ofthe compounds that we discovered. Therefore, smallmolecules identified in whole-organism screens mightbe more relevant than those identified by in vitro andcell-culture-based screens.

Significant questions remain about how closelyzebrafish can model human disease processes andhow reliably zebrafish results translate to humans. Forexample, it is presently unclear how often small mole-cules discovered in zebrafish screens will have similareffects in humans, nor is it clear exactly how predictivezebrafish toxicity will be of human toxicity. However,initial findings indicate a high degree of similaritybetween the drug responses of zebrafish andhumans69. Our studies of many chemicals that affectthe cell cycle show that 50–70% of the chemicalsknown to affect the cell cycle in mammalian cell culturehave similar effects on the zebrafish cell cycle (R.Murphey and L.I.Z, unpublished observations). Otherstudies indicate an even greater degree of conservationof drug responses, with estimates reaching 95%68.These results might reflect the high degree of amino-acid sequence identity between zebrafish and humandrug targets, particularly at protein active sites wheremany drugs bind. Nonetheless, it is important to real-ize that zebrafish screening might lead to the identifi-cation of small molecules that are species specific, per-haps reflecting differences between zebrafish andhuman protein structure or ADME.

In recent years, efforts to improve animal studies indrug development have focused on developing ‘high-end’ animal models that more faithfully mimic humanbiology70. These efforts have included genetic manipu-lation of traditional animal models, including rodents,and the adoption of new large animal species. In manyinstances, large animals and genetically modifiedsmall animals have made prediction of drug efficacyand toxicity more reliable. However, as animal modelsbecome more sophisticated, they also tend to becomemore expensive and are necessarily limited to use in asmaller number of experiments. In the near future, thezebrafish is unlikely to contribute to efforts at morefaithfully modelling human disease and toxicity. Formany applications, primates and other large mammalswill continue to be the gold standard for assessing efficacyand safety in the late stages of drug development.However, the zebrafish offers the ability to test the effi-cacy and safety of thousands of compounds quickly andinexpensively. This approach will not eliminate the needfor sophisticated mammalian assays but will be appro-priate for many circumstances in which cost, scale andefficiency are more important than perfect replication ofhuman physiology. The rapid, inexpensive prediction ofthe biological activity of a small molecule might be espe-cially useful for phenotype-based discovery of novel drugleads and in preliminary assessments of compound

same attributes are beginning to provide opportunitiesfor increases in efficiency in the fields of drug discoveryand development. Genetic and morpholino oligo-nucleotide screens are being used to systematicallyidentify and validate novel therapeutic targets. Zebrafishdisease models are being used in whole-organism screensto identify small molecules that suppress the diseasestate; and zebrafish are proving to be a useful organismfor toxicology, allowing the safety of compounds to beassessed earlier, more quickly and less expensively.

What advantages do zebrafish offer over conventionalin vitro approaches to target validation, lead discoveryand toxicology? First, the whole-organism approachallows biological questions to be addressed that simplycannot be addressed in vitro. For instance, one of ourscreens has been used to identify small molecules thatreverse the cardiac dilation and contractility defects in azebrafish model of heart failure (C. Hong and R.T.P.,unpublished observations). Second, even when an in vitroscreen is available, the results obtained by whole-organ-ism screens might be more relevant than thoseobtained using the in vitro screen. As we have shown incollaboration with the Institute for Chemistry and CellBiology at Harvard Medical School, many of the smallmolecules that we identified in our zebrafish cell-cyclescreen were not identified in analogous in vitro cell-line screens (L.I.Z., unpublished observations). There

DISRUPTIVE TECHNOLOGY

A new technology that is simplerand cheaper than the prevailingtechnology but initially offersreduced performance. As thetechnology’s capabilitiesimprove, its simplicity and costmight allow it to supplant theprevailing technology.

Structure

Compound L

Activity

+

+

+

+

+/–

Structure Activity

+/–

+/–

+/–

–

–

–

Cl

NH

O

O

O

CH3

Cl

NH

O

O

O

CH3

Cl

NH

O O

FF F

OH

NH

O

O

O

CH3

Br

NH

O

OH

Br

O

Br

NH

O

OH

O

H3C

Cl

Cl

NH

O

OH

O

I

NH

O

OH

O

Br

NH

O

OH

O

OMe

NH

O

OH

O

NH

OF

O

O

CH3

Figure 4 | Structure–activity relationship (SAR) studies in zebrafish. Compound L wasidentified in a zebrafish assay for changes in phosphohistone H3 levels. Testing derivatives ofL in zebrafish rapidly identified modifications that were tolerated without loss of efficacy (+),modifications that resulted in a fivefold loss of efficacy (+/–) and modifications that eliminatedefficacy (–). These in vivo results integrate changes in receptor binding and in absorption,distribution, metabolism and excretion (ADME).


R E V I E W S

The zebrafish has already provided a wealth of fun-damental information about embryonic developmentand disease. With the completion of the zebrafishgenome project and the establishment of a robust infra-structure for genetic and physiological studies, thezebrafish system sits poised to take on a larger role in thefield of drug development. By contributing to targetidentification and validation, drug lead discovery andtoxicology, the zebrafish might provide a shorter path todeveloping novel therapies for human disease.

safety. By providing a low-cost and efficient (albeitimperfect) alternative to other animal models, zebrafishchemical screens possess the hallmarks of a DISRUPTIVE

TECHNOLOGY71. Similar to other disruptive technologies,zebrafish chemical screens are likely to allow novelapplications that were previously cost-prohibited andto replace other animal models for simple experi-ments. As technologies improve, the zebrafish mightbe able to replace mammalian models in applicationsof increasing complexity.

1. Eisen, J. S. Zebrafish make a big splash. Cell 87, 969–977(1996).

2. Driever, W. et al. A genetic screen for mutations affectingembryogenesis in zebrafish. Development 123, 37–46(1996).

3. Haffter, P. et al. The identification of genes with unique andessential functions in the development of the zebrafish,Danio rerio. Development 123, 1–36 (1996).References 2 and 3 describe the first large-scalegenetic screens in zebrafish. These screens showedthe remarkable potential of zebrafish screens andfuelled the rapid increase in the use of zebrafish forresearch.

4. Stickney, H. L. et al. Rapid mapping of zebrafish mutationswith SNPs and oligonucleotide microarrays. Genome Res.12, 1929–1934 (2002).

5. Ton, C., Stamatiou, D., Dzau, V. J. & Liew, C. C.Construction of a zebrafish cDNA microarray: geneexpression profiling of the zebrafish during development.Biochem. Biophys. Res. Commun. 296, 1134–1142(2002).

6. Nasevicius, A. & Ekker, S. C. Effective targeted gene‘knockdown’ in zebrafish. Nature Genet. 26, 216–220(2000).This article describes the first use of antisensemorpholino oligonucleotides to knock down genefunction. Morpholino oligonucleotides continue to bethe preferred technique for rapid, reliable geneinactivation in zebrafish.

7. Heasman, J. Morpholino oligos: making sense of antisense?Dev. Biol. 243, 209–214 (2002).

8. Meng, A., Jessen, J. R. & Lin, S. Transgenesis. MethodsCell Biol. 60, 133–148 (1999).

9. Davidson, A. E. et al. Efficient gene delivery and geneexpression in zebrafish using the Sleeping Beautytransposon. Dev. Biol. 263, 191–202 (2003).

10. Kurita, K., Burgess, S. M. & Sakai, N. Transgenic zebrafishproduced by retroviral infection of in vitro-cultured sperm.Proc. Natl Acad. Sci. USA 101, 1263–1267 (2004).

11. Wienholds, E., Schulte-Merker, S., Walderich, B. & Plasterk,R. H. Target-selected inactivation of the zebrafish rag1 gene.Science 297, 99–102 (2002).

12. Lee, K. Y., Huang, H., Ju, B., Yang, Z. & Lin, S. Clonedzebrafish by nuclear transfer from long-term-cultured cells.Nature Biotechnol. 20, 795–799 (2002).

13. Bhatt, D. H., Otto, S. J., Depoister, B. & Fetcho, J. R. CyclicAMP-induced repair of zebrafish spinal circuits. Science305, 254–258 (2004).

14. Farber, S. A. et al. Genetic analysis of digestive physiologyusing fluorescent phospholipid reporters. Science 292,1385–1388 (2001).

15. Lindsay, M. A. Target discovery. Nature Rev. Drug Discov.2, 831–838 (2003).

16. Bleicher, K. H., Bohm, H. J., Muller, K. & Alanine, A. I. Hitand lead generation: beyond high-throughput screening.Nature Rev. Drug Discov. 2, 369–378 (2003).

17. Armer, R. E. & Morris, I. D. Trends in early drug safety. DrugNews Perspect. 17, 143–148 (2004).

18. Shin, J. T. & Fishman, M. C. From zebrafish to human:modular medical models. Annu. Rev. Genomics Hum.Genet. 3, 311–340 (2002).

19. Rubinstein, A. L. Zebrafish: from disease modeling to drugdiscovery. Curr. Opin. Drug Discov. Devel. 6, 218–223 (2003).

20. Amatruda, J. F., Shepard, J. L., Stern, H. M. & Zon, L. I.Zebrafish as a cancer model system. Cancer Cell 1,229–231 (2002).

21. Amsterdam, A. et al. Identification of 315 genes essential forearly zebrafish development. Proc. Natl Acad. Sci. USA101, 12792–12797 (2004).

22. Sun, Z. et al. A genetic screen in zebrafish identifies ciliagenes as a principal cause of cystic kidney. Development131, 4085–4093 (2004).

23. Otto, E. A. et al. Mutations in INVS encoding inversin causenephronophthisis type 2, linking renal cystic disease to thefunction of primary cilia and left-right axis determination.Nature Genet. 34, 413–420 (2003).

24. Poss, K. D., Keating, M. T. & Nechiporuk, A. Tales ofregeneration in zebrafish. Dev. Dyn. 226, 202–210 (2003).

25. Gerull, B. et al. Mutations of TTN, encoding the giant musclefilament titin, cause familial dilated cardiomyopathy. NatureGenet. 30, 201–204 (2002).

26. Xu, X. et al. Cardiomyopathy in zebrafish due to mutation inan alternatively spliced exon of titin. Nature Genet. 30,205–209 (2002).

27. Li, Q. Y. et al. Holt-Oram syndrome is caused by mutationsin TBX5, a member of the Brachyury (T) gene family. NatureGenet. 15, 21–29 (1997).

28. Garrity, D. M., Childs, S. & Fishman, M. C. The heartstringsmutation in zebrafish causes heart/fin Tbx5 deficiencysyndrome. Development 129, 4635–4645 (2002).

29. Curran, M. E. et al. A molecular basis for cardiac arrhythmia:HERG mutations cause long QT syndrome. Cell 80,795–803 (1995).

30. Langheinrich, U., Vacun, G. & Wagner, T. Zebrafish embryosexpress an orthologue of HERG and are sensitive toward arange of QT-prolonging drugs inducing severe arrhythmia.Toxicol. Appl. Pharmacol. 193, 370–382 (2003).

31. Donovan, A. et al. Positional cloning of zebrafish ferroportin1identifies a conserved vertebrate iron exporter. Nature 403,776–781 (2000).This article shows that the hypochromic anaemiaphenotype of the zebrafish mutant weissherbst iscaused by mutation of ferroportin 1. This is the firsttime that a zebrafish gene has led to the discovery ofa new human disease gene.

32. Njajou, O. T. et al. A mutation in SLC11A3 is associated withautosomal dominant hemochromatosis. Nature Genet. 28,213–214 (2001).

33. Davidson, A. J. et al. cdx4 mutants fail to specify bloodprogenitors and can be rescued by multiple hox genes.Nature 425, 300–306 (2003).

34. Kalev-Zylinska, M. L. et al. Runx1 is required for zebrafishblood and vessel development and expression of a humanRUNX1-CBF2T1 transgene advances a model for studies ofleukemogenesis. Development 129, 2015–2030 (2002).

35. Langenau, D. M. et al. Myc-induced T cell leukemia intransgenic zebrafish. Science 299, 887–890 (2003).

36. Li, L. & Dowling, J. E. A dominant form of inherited retinaldegeneration caused by a non-photoreceptor cell-specificmutation. Proc. Natl Acad. Sci. USA 94, 11645–11650 (1997).

37. Peitsaro, N., Kaslin, J., Anichtchik, O. V. & Panula, P.Modulation of the histaminergic system and behaviour byα-fluoromethylhistidine in zebrafish. J. Neurochem. 86,432–441 (2003).

38. Nicolson, T. et al. Genetic analysis of vertebrate sensory haircell mechanosensation: the zebrafish circler mutants.Neuron 20, 271–283 (1998).

39. Malicki, J. et al. Mutations affecting development of thezebrafish ear. Development 123, 275–283 (1996).

40. Neuhauss, S. C. et al. Genetic disorders of vision revealedby a behavioral screen of 400 essential loci in zebrafish. J. Neurosci. 19, 8603–8615 (1999).

41. Darland, T. & Dowling, J. E. Behavioral screening for cocainesensitivity in mutagenized zebrafish. Proc. Natl Acad. Sci.USA 98, 11691–11696 (2001).

42. Carpinelli, M. R. et al. Suppressor screen in Mpl–/– mice: c-Myb mutation causes supraphysiological production ofplatelets in the absence of thrombopoietin signaling. Proc.Natl Acad. Sci. USA 101, 6553–6558 (2004).In this study, a genetic screen identified mutations inc-Myb that can suppress the disease phenotype in amurine model of thrombocytopaenia. This articleshows the feasibility of using genetic screens toidentify novel drug targets.

43. Kenwrick, S., Amaya, E. & Papalopulu, N. Pilotmorpholino screen in Xenopus tropicalis identifies a novelgene involved in head development. Dev. Dyn. 229,289–299 (2004).

44. Yamada, L. et al. Morpholino-based gene knockdownscreen of novel genes with developmental function in Cionaintestinalis. Development 130, 6485–6495 (2003).

45. Tobert, J. A. Lovastatin and beyond: the history of theHMG-CoA reductase inhibitors. Nature Rev. Drug Discov.2, 517–526 (2003).

46. Clader, J. W. The discovery of ezetimibe: a view fromoutside the receptor. J. Med. Chem. 47, 1–9 (2004).

47. Altmann, S. W. et al. Niemann-Pick C1 Like 1 protein iscritical for intestinal cholesterol absorption. Science 303,1201–1204 (2004).

48. Robson, J. M. & Keele, C. A. Recent Advances inPharmacology (J. & A. Churchill, London 1956).

49. Peterson, R. T. et al. Chemical suppression of a geneticmutation in a zebrafish model of aortic coarctation. NatureBiotechnol. 22, 595–599 (2004).This article describes the discovery of compoundsthat completely suppress the phenotypic effects of azebrafish vascular mutation, revealing the potentialof zebrafish screens for discovering disease-suppressing compounds.

50. Stern, H. M. & Zon, L. I. Cancer genetics and drug discoveryin the zebrafish. Nature Rev. Cancer 3, 533–539 (2003).

51. Zhong, T. P., Rosenberg, M., Mohideen, M. A., Weinstein, B.& Fishman, M. C. gridlock, an HLH gene required forassembly of the aorta in zebrafish. Science 287, 1820–1824(2000).

52. Zhong, T. P., Childs, S., Leu, J. P. & Fishman, M. C. Gridlocksignalling pathway fashions the first embryonic artery. Nature414, 216–220 (2001).

53. Langenau, D. M. et al. In vivo tracking of T cell development,ablation, and engraftment in transgenic zebrafish. Proc. NatlAcad. Sci. USA 101, 7369–7374 (2004).

54. Traver, D. et al. Effects of lethal irradiation in zebrafish andrescue by hematopoietic cell transplantation. Blood 104,1298–1305 (2004).

55. Poss, K. D., Wilson, L. G. & Keating, M. T. Heartregeneration in zebrafish. Science 298, 2188–2190 (2002).

56. Guo, S. Linking genes to brain, behavior and neurologicaldiseases: what can we learn from zebrafish? Genes BrainBehav. 3, 63–74 (2004).

57. Ulrich, R. & Friend, S. H. Toxicogenomics and drugdiscovery: will new technologies help us produce betterdrugs? Nature Rev. Drug Discov. 1, 84–88 (2002).

58. Llorens, O., Perez, J. J. & Villar, H. O. Toward the design ofchemical libraries for mass screening biased againstmutagenic compounds. J. Med. Chem. 44, 2793–2804(2001).

59. Pritchard, J. F. et al. Making better drugs: decision gates innon-clinical drug development. Nature Rev. Drug Discov. 2,542–553 (2003).

60. Spitsbergen, J. M. & Kent, M. L. The state of the art of thezebrafish model for toxicology and toxicologic pathologyresearch — advantages and current limitations. Toxicol.Pathol. 31 Suppl, 62–87 (2003).

61. Parng, C., Seng, W. L., Semino, C. & McGrath, P. Zebrafish:a preclinical model for drug screening. Assay Drug Dev.Technol. 1, 41–48 (2002).

62. Brion, F. et al. Impacts of 17β-estradiol, includingenvironmentally relevant concentrations, on reproductionafter exposure during embryo-larval-, juvenile- and adult-lifestages in zebrafish (Danio rerio). Aquat. Toxicol. 68,193–217 (2004).

63. Levin, E. D., Chrysanthis, E., Yacisin, K. & Linney, E.Chlorpyrifos exposure of developing zebrafish: effects onsurvival and long-term effects on response latency andspatial discrimination. Neurotoxicol. Teratol. 25, 51–57(2003).


R E V I E W S

64. Schilling, T. F. & Knight, R. D. Origins of anteroposteriorpatterning and Hox gene regulation during chordateevolution. Philos. Trans. R Soc. Lond. B Biol. Sci. 356,1599–1613 (2001).

65. Spitsbergen, J. M. et al. Neoplasia in zebrafish (Danio rerio)treated with 7,12-dimethylbenz[a]anthracene by twoexposure routes at different developmental stages. Toxicol.Pathol. 28, 705–715 (2000).

66. Murakami, S. L. et al. Developmental differences insusceptibility to neomycin-induced hair cell death in thelateral line neuromasts of zebrafish (Danio rerio). Hear. Res.186, 47–56 (2003).

67. Amanuma, K., Takeda, H., Amanuma, H. & Aoki, Y.Transgenic zebrafish for detecting mutations caused bycompounds in aquatic environments. Nature Biotechnol.18, 62–65 (2000).

68. Milan, D. J., Peterson, T. A., Ruskin, J. N., Peterson, R. T. &MacRae, C. A. Drugs that induce repolarizationabnormalities cause bradycardia in zebrafish. Circulation107, 1355–1358 (2003).In this study, the cardiotoxicity of 100 compoundswas tested using an automated zebrafish assay. A high degree of correlation was observed betweenzebrafish toxicity and known human toxicities,indicating that zebrafish might be well suited forstudies of cardiotoxicity.

69. Langheinrich, U. Zebrafish: a new model on thepharmaceutical catwalk. Bioessays 25, 904–912 (2003).

70. Greaves, P., Williams, A. & Eve, M. First dose of potential newmedicines to humans: how animals help. Nature Rev. DrugDiscov. 3, 226–236 (2004).

71. Christensen, C. M. The Innovator’s Dilemma: When NewTechnologies Cause Great Firms to Fail (Harvard BusinessSchool Press, Boston, Massachusetts, 1997).

72. Solnica-Krezel, L., Schier, A. F. & Driever, W. Efficient recoveryof ENU-induced mutations from the zebrafish germline.Genetics 136, 1401–1420 (1994).

73. Amsterdam, A. Insertional mutagenesis in zebrafish. Dev. Dyn.228, 523–534 (2003).

74. Kwok, C. et al. Characterization of whole genome radiationhybrid mapping resources for non-mammalian vertebrates.Nucleic Acids Res. 26, 3562–3566 (1998).

75. Hukriede, N. A. et al. Radiation hybrid mapping of the zebrafishgenome. Proc. Natl Acad. Sci. USA 96, 9745–9750 (1999).

76. Geisler, R. et al. A radiation hybrid map of the zebrafishgenome. Nature Genet. 23, 86–89 (1999).

77. Affymetrix GeneChip Array Products [online],<http://www.affymetrix.com/products/arrays/>

78. Zebrafish Genome Resources [online],<http://www.ncbi.nlm.nih.gov/genome/guide/zebrafish/>

79. Zebrafish International Resource Center [online],<http://zfin.org/zirc/home/guide.php>

80. Tübingen Zebrafish Stockcenter [online],<http://www.eb.tuebingen.mpg.de/services/stockcenter/>

81. The Sanger Institute Danio rerio Sequencing Project [online],<http://www.sanger.ac.uk/Projects/D_rerio/>

82. Trans-NIH Zebrafish Initiative [online],<http://www.nih.gov/science/models/zebrafish/>

Competing interests statementThe authors declare competing financial interests: see Web versionfor details.

Online links

DATABASESThe following terms in this article are linked online to:Entrez Gene:http://www.ncbi.nih.gov/entrez/query.fcgi?db=genefli 1 | hey2 | KCNH2 | pkd2 | TBX5 | tcf2 | TTN | vegfOMIM: http://www.ncbi.nlm.nih.gov/Omim/chronic myelogenous leukaemia | polycystic kidney disease

FURTHER INFORMATIONFDA report: ‘Innovation or stagnation: challenge andopportunity on the critical path to new medical products’:www.fda.gov/oc/initiatives/criticalpath/whitepaper.htmlAccess to this interactive links box is free online.

in vivo drug discovery in the zebrafish

Documents